PeterCxy/Step-3.7-Flash-ARA-Heretic-GGUF
Quantization (fewer layers) vs PeterCxy/Step-3.7-Flash-ARA-Heretic. PeterCxy/Step-3.7-Flash-ARA-Heretic-GGUF matches part of PeterCxy/Step-3.7-Flash-ARA-Heretic (711 of its 1420 tensors; the rest of PeterCxy/Step-3.7-Flash-ARA-Heretic has no counterpart). On the shared part: PeterCxy/Step-3.7-Flash-ARA-Heretic-GGUF is a GGUF quantization (Q8_0) of the same weights: 223 tensors match exactly and 488 within quantization error.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/181, MLP 0/261, Attention 0/225, other 0/42